Filmora Video Enhancer

Cloud-based video enhancement skill that uploads a user-selected local video file to Wondershare-hosted endpoints for AI enhancement, then downloads the proc...

Security Scan

The skill's code, install spec, and runtime instructions are consistent with a cloud-based video enhancement workflow that uploads a user-chosen file to Wondershare endpoints; nothing in the bundle indicates hidden or unrelated behavior, although a few privacy and minor consistency issues should be noted before installing.

Purpose & Capability

The skill is described as a cloud video enhancer and the bundled script, declared brew install of ffmpeg, and network endpoints all align with that purpose. Minor inconsistency: the SKILL.md and script use/require ffprobe for metadata inspection, but the manifest's required binaries list only 'ffmpeg' (ffprobe is a separate binary provided by the ffmpeg package on many systems but should be explicitly listed).

Instruction Scope

SKILL.md instructs the agent to ask for explicit consent, validate a single local video file, read metadata via ffprobe, upload that file and limited metadata to specific Wondershare endpoints, poll for completion, and download the result. The instructions explicitly forbid scanning directories, modifying unrelated files, or uploading anything other than the chosen file; the script's visible logic follows that flow.

Install Mechanism

The only install action is a Homebrew formula to install ffmpeg, which is a standard and well-known package distribution method. No arbitrary downloads, third-party installers, or extracted archives are used.

Credentials

No credentials or environment variables are requested, which is proportionate. However the workflow collects and transmits metadata that may be privacy-sensitive: file MD5, duration, resolution, and a device identifier derived from hostname and MAC (hashed with SHA-256). These behaviors are declared in SKILL.md but are privacy-relevant and should be accepted explicitly by the user before use.

Persistence & Privilege

The skill is not always-enabled, does not request persistent elevated privileges, and does not modify other skills or system-wide agent settings. It runs as an on-demand script and requires explicit user invocation/consent per SKILL.md.

Assessment

This skill appears to do what it says: upload a single local .mp4/.mov file to Filmora/Wondershare cloud endpoints for AI enhancement and download the result. Before installing or running it, confirm the user explicitly consents to third‑party cloud processing and understands privacy implications (the tool sends file MD5, duration/size, and a SHA‑256 device identifier derived from hostname+MAC). Ensure ffprobe is available on PATH (the manifest mentions ffmpeg but ffprobe is used for metadata), and avoid using this on sensitive or confidential videos. If you want extra caution, (1) inspect the full script for any additional endpoints or code paths, (2) run it in an isolated environment or VM, and (3) monitor network traffic during a test run to verify it only contacts the documented https://filmora-cloud-api-alisz.wondershare.cc endpoints and the task result download URL.

Video Enhancer

Enhance a user-selected local video using Wondershare-hosted cloud processing.

This is a cloud workflow, not a fully local or offline enhancement pipeline.

Use This Skill When

Use this skill when the user wants to:

• Enhance a blurry or low-quality local video
• Upscale a local .mp4 or .mov video
• Convert a local video to a clearer HD result using cloud processing

Do not use this skill when:

• The user wants a fully local or offline workflow
• The task is trimming, editing, subtitle work, or format conversion only
• The input is a remote URL instead of a local file
• The user does not accept third-party cloud upload

Preconditions

Before running:

• Confirm the input file exists
• Confirm the file extension is .mp4 or .mov
• Confirm ffprobe is available
• ⚠️ Tell the user this workflow uploads the source file to a third-party cloud service operated by Wondershare
• ⚠️ Use the skill only after the user accepts cloud processing
• ⚠️ The file is not sensitive, private, or confidential
• Enforce these limits:
  • If max(width, height) <= 1920, duration must be <= 300 seconds
  • If max(width, height) > 1920, duration must be <= 60 seconds
• Refuse the request if any limit fails and explain which limit was exceeded

Privacy and Network Disclosure

This workflow uploads the local video file and basic metadata to Wondershare-operated cloud endpoints.

Uploaded data includes:

• The source video file
• Video duration
• Video width and height
• File MD5 checksum
• Computer name and MAC address to generate a non‑reversible SHA‑256 device ID. Raw data is never uploaded or stored, used only for device identification

Network endpoints used by the bundled script:

• https://filmora-cloud-api-alisz.wondershare.cc/open/v1/resources/upload
• https://filmora-cloud-api-alisz.wondershare.cc/open/v1/tasks
• https://filmora-cloud-api-alisz.wondershare.cc/open/v1/tasks/<task_id>
• The final download URL returned by the task result

Do not use this skill for sensitive files unless the user explicitly accepts third-party processing.

Permissions

This skill requires:

• Read access only to the user-specified input video file
• Write access only to the selected output directory, or the input file's parent directory if no output directory is provided
• Network access to the Wondershare-hosted endpoints listed above
• Local execution of the fixed ffprobe binary for metadata inspection

This skill must not:

• Modify the original input file in place
• Scan directories for other media files
• Upload any file other than the user-specified input file
• Write arbitrary files outside the chosen output location

Preferred Execution

Run the script using an absolute path:

python {baseDir}/scripts/video_enhance.py -i "/absolute/path/to/video.mp4" -o "/output/dir"

If the output directory is omitted, the script saves the enhanced file next to the input video.

Workflow

1. Validate that the input file exists and has a supported extension
2. Read the local video metadata with ffprobe
3. Enforce duration and resolution limits
4. Upload the video file and metadata to the Wondershare cloud API
5. Submit the enhancement task
6. Poll until the task succeeds or fails
7. Download the enhanced result to the output directory
8. Report the saved file path to the user

Input and Output

Input:

• Local video path
• Optional output directory

Output:

• Enhanced video saved to disk
• Filename pattern:

<input_stem>_hd_YYYYMMDD_HHMMSS.<ext>

Supported Formats

Supported input formats:

• .mp4
• .mov

Security Notes

• This is a third-party cloud processing workflow
• The bundled script invokes only the fixed ffprobe binary with predefined arguments for local metadata inspection
• The workflow does not require arbitrary shell execution
• The workflow should use only the declared Wondershare-hosted endpoints and the final HTTPS task result download URL
• The tool should be treated as unsuitable for sensitive or confidential media

Cloud Processing

• Videos are uploaded to filmora cloud services for AI enhancement
• Cloud providers temporarily store videos during processing
• Enhanced videos are downloaded to local storage
• Cloud-side data is deleted after processing completes

Troubleshooting

Check that the file exists:

ls -lh "/absolute/path/to/video.mp4"

Check that ffprobe is available:

ffprobe -version

If the task fails:

• Verify network connectivity
• Re-run the script and inspect the error output
• Confirm the file is within the documented duration and format limits

Script Location

Bundled script:

• {baseDir}/scripts/video_enhance.py

Agent Guidance

• Always tell the user that the selected local video will be uploaded to a Wondershare-operated third-party cloud service before running the workflow
• Require explicit user consent in the current conversation before execution
• Refuse sensitive, private, or confidential videos
• Keep the response focused on the saved file path, cloud-processing disclosure, and any relevant warnings